Ambiguity attitudes and surprises

5.1 Setup

5.1.1 Install Packages

We install the following packages using the groundhog package manager to increase computational reproducibility.

options(repos = c(CRAN = "https://cran.r-project.org")) 

if (!requireNamespace("groundhog", quietly = TRUE)) {
    install.packages("groundhog")
}

pkgs <- c("magrittr", "data.table", "stringr", "Rmisc", "ggplot2",
          "lmtest", "sandwich", "stargazer")

groundhog::groundhog.library(pkg = pkgs,
                             date = "2024-08-01")

rm(pkgs)

5.1.2 Read Data

data      <- readRDS(file="../data/processed/full.Rda")
timeSpent <- data.table::fread(file = "../data/raw/PageTimes-2021-09-15.csv")
raw       <- data.table::fread(file="../data/raw/all_apps_wide_2021-09-15.csv")

5.2 Table B.1

::: {#tbl-B-1 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E1 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E1 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E1 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E1 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E1 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E1 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E1",
            dep.var.caption = "Dependent variable: E1",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E1**

	Dependent variable: E1


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	-0.962			-1.492	0.951	-2.435
	(1.080)			(1.760)	(1.993)	(1.855)

both		2.430	1.487
		(1.759)	(1.856)

interval		1.331	3.773^**
		(1.836)	(1.923)

stage 2	-1.036	0.955	7.468^***	0.955	-1.634	-2.504^*
	(0.846)	(1.500)	(1.791)	(1.500)	(1.498)	(1.391)

contradiction x stage 2	7.563^***			6.513^***	4.269^*	12.098^***
	(1.315)			(2.336)	(2.232)	(2.251)

both x stage 2		-3.459^*	2.126
		(2.046)	(2.517)

interval x stage 2		-2.589	-4.833^**
		(2.120)	(2.437)

Constant	47.893^***	46.653^***	45.161^***	46.653^***	47.984^***	49.083^***
	(0.747)	(1.193)	(1.294)	(1.193)	(1.395)	(1.293)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.014	0.002	0.024	0.015	0.006	0.028
Adjusted R²	0.013	-0.002	0.021	0.012	0.003	0.025
Residual Std. Error	23.210	22.727	23.652	22.494	23.624	23.473
F Statistic	13.967^***	0.530	7.531^***	5.226^***	2.152^*	9.393^***

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

5.3 Table B.2

::: {#tbl-B-2 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E2 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E2 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E2 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E2 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E2 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E2 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E2",
            dep.var.caption = "Dependent variable: E2",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E2**

	Dependent variable: E2


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	-1.235			-1.497	-1.578	-0.612
	(1.072)			(1.779)	(1.943)	(1.852)

both		1.558	2.443
		(1.806)	(1.825)

interval		0.174	0.092
		(1.867)	(1.858)

stage 2	3.535^***	4.227^***	-3.103^**	4.227^***	2.998^**	3.348^**
	(0.782)	(1.159)	(1.407)	(1.159)	(1.468)	(1.433)

contradiction x stage 2	-5.636^***			-7.331^***	-3.812^*	-5.768^***
	(1.125)			(1.824)	(1.977)	(2.059)

both x stage 2		-0.879	0.683
		(1.843)	(2.041)

interval x stage 2		-1.230	2.289
		(1.871)	(1.932)

Constant	50.681^***	50.108^***	48.611^***	50.108^***	50.282^***	51.666^***
	(0.757)	(1.250)	(1.265)	(1.250)	(1.387)	(1.304)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.013	0.007	0.005	0.023	0.008	0.010
Adjusted R²	0.012	0.004	0.002	0.020	0.005	0.007
Residual Std. Error	21.997	22.066	21.941	20.890	22.813	22.276
F Statistic	12.879^***	2.214^*	1.655	7.827^***	2.700^**	3.462^**

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

5.4 Table B.3

::: {#tbl-B-3 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E3 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E3 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E3 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E3 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E3 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E3 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E3",
            dep.var.caption = "Dependent variable: E3",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E3**

	Dependent variable: E3


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	-0.317			1.231	0.356	-2.623
	(1.049)			(1.814)	(1.864)	(1.767)

both		4.813^***	0.958
		(1.752)	(1.828)

interval		2.199	1.323
		(1.787)	(1.889)

stage 2	0.148	2.161	-5.425^***	2.161	-0.006	-1.779
	(0.818)	(1.477)	(1.635)	(1.477)	(1.283)	(1.469)

contradiction x stage 2	-3.721^***			-7.585^***	-2.558	-0.971
	(1.217)			(2.204)	(1.972)	(2.136)

both x stage 2		-3.940^*	2.675
		(2.083)	(2.253)

interval x stage 2		-2.167	2.861
		(1.956)	(2.217)

Constant	48.591^***	46.278^***	47.510^***	46.278^***	48.477^***	51.091^***
	(0.733)	(1.204)	(1.357)	(1.204)	(1.321)	(1.274)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.006	0.004	0.009	0.012	0.002	0.007
Adjusted R²	0.005	0.001	0.006	0.009	-0.001	0.004
Residual Std. Error	22.475	21.461	23.399	22.160	22.847	22.373
F Statistic	5.563^***	1.320	2.815^**	4.050^***	0.678	2.489^*

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

5.5 Table B.4

::: {#tbl-B-4 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E12 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E12 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E12 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E12 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E12 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E12 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E12",
            dep.var.caption = "Dependent variable: E12",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E12**

	Dependent variable: E12


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	-0.453			0.446	-1.087	-0.776
	(1.031)			(1.772)	(1.754)	(1.838)

both		1.672	0.451
		(1.760)	(1.850)

interval		2.072	0.539
		(1.743)	(1.783)

stage 2	1.817^**	3.090^**	4.460^***	3.090^**	-0.401	2.684^*
	(0.828)	(1.365)	(1.611)	(1.365)	(1.499)	(1.440)

contradiction x stage 2	3.950^***			1.370	5.012^**	5.592^***
	(1.198)			(2.112)	(2.050)	(2.065)

both x stage 2		-0.406	3.816^*
		(1.984)	(2.187)

interval x stage 2		-3.491^*	0.150
		(2.027)	(2.134)

Constant	58.358^***	57.127^***	57.573^***	57.127^***	59.200^***	58.800^***
	(0.716)	(1.230)	(1.276)	(1.230)	(1.234)	(1.258)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.010	0.004	0.019	0.008	0.007	0.021
Adjusted R²	0.009	0.001	0.015	0.005	0.004	0.018
Residual Std. Error	22.155	21.169	23.078	22.444	21.919	22.090
F Statistic	10.599^***	1.214	5.751^***	2.679^**	2.266^*	7.149^***

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

5.6 Table B.5

::: {#tbl-B-5 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E13 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E13 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E13 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E13 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E13 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E13 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E13",
            dep.var.caption = "Dependent variable: E13",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E13**

	Dependent variable: E13


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	0.922			-2.686	4.258^**	1.173
	(1.037)			(1.743)	(1.846)	(1.792)

both		0.479	4.338^**
		(1.749)	(1.786)

interval		-1.358	5.586^***
		(1.811)	(1.779)

stage 2	-1.549^*	-1.333	4.212^***	-1.333	0.068	-3.362^**
	(0.873)	(1.414)	(1.579)	(1.414)	(1.581)	(1.543)

contradiction x stage 2	3.721^***			5.546^***	-0.192	5.848^***
	(1.221)			(2.120)	(2.054)	(2.176)

both x stage 2		-2.029	-1.726
		(2.093)	(2.201)

interval x stage 2		1.401	-4.336^**
		(2.121)	(2.052)

Constant	55.130^***	55.414^***	52.728^***	55.414^***	54.056^***	55.893^***
	(0.736)	(1.200)	(1.264)	(1.200)	(1.357)	(1.273)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.006	0.002	0.010	0.005	0.009	0.013
Adjusted R²	0.005	-0.001	0.007	0.002	0.006	0.010
Residual Std. Error	21.788	22.070	21.473	21.291	21.762	22.259
F Statistic	5.977^***	0.717	3.041^***	1.812	3.054^**	4.266^***

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

5.7 Table B.6

::: {#tbl-B-6 .cell tbl-cap=’ ’}

ols_1 <- lm(formula = E23 ~ surprise + treated + surprise * treated,
              data = data)
  se_1  <- coeftest(x = ols_1, 
                    vcov = vcovCL(ols_1,
                                  cluster = ~data$participant.label,
                                  type = "HC1"))
  
  ols_2 <- lm(formula = E23 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == FALSE))
  se_2  <- coeftest(x = ols_2, 
                    vcov = vcovCL(ols_2,
                                  cluster = data[surprise == FALSE, participant.label],
                                  type = "HC1"))
  
  ols_3 <- lm(formula = E23 ~ communication + treated + communication * treated, 
              data = data,
              subset = (surprise == TRUE))
  se_3  <- coeftest(x = ols_3, 
                    vcov = vcovCL(ols_3,
                                  cluster = data[surprise == TRUE, participant.label],
                                  type = "HC1"))
  
  ols_4 <- lm(formula = E23 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "point"))
  se_4  <- coeftest(x = ols_4, 
                    vcov = vcovCL(ols_4,
                                  cluster = data[communication == "point", participant.label],
                                  type = "HC1"))
  
  ols_5 <- lm(formula = E23 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "interval"))
  se_5  <- coeftest(x = ols_5, 
                    vcov = vcovCL(ols_5,
                                  cluster = data[communication == "interval", participant.label],
                                  type = "HC1"))
  
  ols_6 <- lm(formula = E23 ~ surprise + treated + surprise * treated, 
              data = data,
              subset = (communication == "both"))
  se_6  <- coeftest(x = ols_6, 
                    vcov = vcovCL(ols_6,
                                  cluster = data[communication == "both", participant.label],
                                  type = "HC1"))
  
  
  
  se <- list(se_1[,2], se_2[,2], se_3[,2], se_4[,2], se_5[,2], se_6[,2])
  p  <- list(se_1[,4], se_2[,4], se_3[,4], se_4[,4], se_5[,4], se_6[,4])
  
  stargazer(ols_1, ols_2, ols_3, ols_4, ols_5, ols_6, 
            align = TRUE, 
            se = se, 
            p = p,   
            title = "Linear regressions: Treatment effects on E23",
            dep.var.caption = "Dependent variable: E23",
            dep.var.labels = " ",
            model.names = FALSE,
            column.labels = c("full", "confirmation", "contradiction", "point", "interval", "both"),
            covariate.labels = c("contradiction", "both", "interval", "stage 2", "contradiction x stage 2", "both x stage 2", "interval x stage 2", "Constant"),
            font.size = "scriptsize",
            type = "html", 
            df = FALSE,
          notes = c("The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)"),
          notes.align = "l")

**Linear regressions: Treatment effects on E23**

	Dependent variable: E23


	full	confirmation	contradiction	point	interval	both
	(1)	(2)	(3)	(4)	(5)	(6)

contradiction	1.513			0.254	1.193	3.086
	(1.154)			(2.075)	(1.963)	(1.959)

both		0.306	3.139
		(2.028)	(2.006)

interval		0.722	1.661
		(2.047)	(1.992)

stage 2	3.099^***	3.308^**	-7.562^***	3.308^**	2.074	3.893^**
	(0.855)	(1.389)	(1.812)	(1.389)	(1.458)	(1.597)

contradiction x stage 2	-10.200^***			-10.869^***	-7.545^***	-12.211^***
	(1.326)			(2.283)	(2.198)	(2.417)

both x stage 2		0.585	-0.756
		(2.116)	(2.564)

interval x stage 2		-1.234	2.091
		(2.013)	(2.447)

Constant	59.658^***	59.322^***	59.575^***	59.322^***	60.043^***	59.628^***
	(0.829)	(1.452)	(1.481)	(1.453)	(1.442)	(1.416)


Observations	3,010	1,490	1,520	1,014	1,002	994
R²	0.020	0.005	0.027	0.026	0.012	0.024
Adjusted R²	0.019	0.001	0.023	0.023	0.009	0.021
Residual Std. Error	23.195	23.236	23.167	23.966	22.547	23.058
F Statistic	20.085^***	1.441	8.284^***	8.939^***	3.966^***	8.016^***

Note:	p<0.1; p<0.05; p<0.01
	The underlying standard errors (“HC1”, in parentheses) are clustered at the individual level and estimated with the R package sandwich (Zeileis 2004; Zeileis et al. 2020)

:::

Session Info

sessionInfo()

R version 4.4.1 (2024-06-14)
Platform: x86_64-apple-darwin20
Running under: macOS Sonoma 14.4.1

Matrix products: default
BLAS:   /Library/Frameworks/R.framework/Versions/4.4-x86_64/Resources/lib/libRblas.0.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-x86_64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: Europe/Zurich
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] stargazer_5.2.3   sandwich_3.1-0    lmtest_0.9-40     zoo_1.8-12       
 [5] ggplot2_3.5.1     Rmisc_1.5.1       plyr_1.8.9        lattice_0.22-6   
 [9] stringr_1.5.1     data.table_1.15.4 magrittr_2.0.3   

loaded via a namespace (and not attached):
 [1] gtable_0.3.5      jsonlite_1.8.8    dplyr_1.1.4       compiler_4.4.1   
 [5] tidyselect_1.2.1  Rcpp_1.0.13       parallel_4.4.1    scales_1.3.0     
 [9] yaml_2.3.10       fastmap_1.2.0     R6_2.5.1          generics_0.1.3   
[13] knitr_1.48        htmlwidgets_1.6.4 tibble_3.2.1      munsell_0.5.1    
[17] pillar_1.9.0      rlang_1.1.4       utf8_1.2.4        stringi_1.8.4    
[21] xfun_0.46         cli_3.6.3         withr_3.0.1       digest_0.6.36    
[25] grid_4.4.1        rstudioapi_0.16.0 lifecycle_1.0.4   vctrs_0.6.5      
[29] evaluate_0.24.0   glue_1.7.0        groundhog_3.2.0   fansi_1.0.6      
[33] colorspace_2.1-1  rmarkdown_2.27    tools_4.4.1       pkgconfig_2.0.3  
[37] htmltools_0.5.8.1